CPU Boards - performance comparison

Please find below some performance comparison we've done for a set of most popular mini cpu-boards evaluated for OGN receivers.
In our tests we used sysbench in different configurations.

Please note the results are ordered from most-performant (left) to least performant (right).

Benchmarks

ODROID-XU4 (Exynos5422 Cortex-A15 2Ghz and Cortex-A7 8-cores) ODROID-U3 (Exynos4412 Prime Cortex-A9 4-cores) Raspberry Pi3 (1.2GHz 64-bit 4-core ARM Cortex-A53 (~10x Pi1 perf)) Rockchip RK3188 (e.g. MK809 IV 4-cores) Odroid-C1 (ARM Cortex-A5 4-cores) OrangePi One (Allwinner H3 Cortex A7 4-cores) Raspberry Pi2 (900MHz 4-core ARM Cortex-A7 CPU (~6x Pi1 perf)) Olimex A20-OLinuXino-LIME (ARM Cortex-A7 2-cores) Banana Pi (ARM Cortex-A7 2-cores) Cubieboard3 (ARM Cortex-A7 2-cores) Cubieboard2 (ARM Cortex-A7, 2-cores) BeagleBone Black (ARM Cortex-A7, 1-core) Raspberry Pi Zero W(ARM1176JZF-S (ARMv6k) 1-core) BeagleBone (ARM Cortex-A8, 1-core) Raspberry Pi (ARM1176JZF-S (ARMv6k) 1-core)
Price 74$ (http://hardkernel.com + 19$ for shipping!)
80€ (http://pollin.de)
65$ (http://hardkernel.com + 25$ for shipping!)
70€ (http://pollin.de)
~43$
39€ (http://pollin.de)
~44-50$ (ebay)
45€ (http://amazon.de)
44€ (http://pollin.de) $9.99 + ~$4 shipping
(aliexpress)
~40$
37€ (http://pollin.de)
33€ (https://www.olimex.com)
40€ (http://exp-tech.de)
~40€, 75€ (kit) (amazon)
35€ (http://redcoon.de)
88€ (http://pollin.de) ~68$
63€ (http://pollin.de) _
58€(ebay, free shipment)
48€ (http://pollin.de) 11€ (https://www.kiwi-electronics.nl) 78€ (http://rs-components.com) ~48$
30€ (http://pollin.de)
CPU
odroid-xu4.png odroid-u3.png r-pi3modelB.jpg rk3188.png odroid-c1.png Orange-Pi-One.jpg r-pi2modelB.png A20-OLinuXino-LIME-1.jpg banana-pi.png Cubieboard_3.png cubie2.png Beaglebone_Black.png raspberry_zero_w.jpeg Beaglebone_White.png r-pi.png
1 Thread 8.7536s 13.9420s 19.6445s 15.1796s 23.8167s 24.8021s 31.0962s 28.6889s 29.5346s 29.0845s 32.6901s 29.4214s 37.3745s 40.9213s 54.0791s
2 Threads 4.3783s 6.9756s 9.8471s 8.4593s 11.5903s 12.4252s 15.6131s 14.4584s 14.4951s 14.6567s 20.4404s 29.3900s 37.4377s 40.7818s 54.2914s
4 Threads 2.1944s 3.4863s 4.9428s 5.0010s 5.8368s 6.2781s 7.8286s 14.4678s 14.5026s 14.7216s 20.3010s 29.3866s 37.4271s 40.8213s 54.2213s
MEMORY
256M 5109.55 MB/sec 3567.99 MB/sec 1531.52 MB/sec 1908.11 MB/sec 1448.41 MB/sec 1075.59 MB/sec 1045.95 MB/sec 1253.92 MB/sec 1147.46 MB/sec 1131.00 MB/sec 1027.03 MB/sec 560.11 MB/sec 405.59 MB/sec 564.73 MB/sec 352.03 MB/sec
512M 5132.19 MB/sec 4192.24 MB/sec 1928.87 MB/sec 2022.42 MB/sec 1445.04 MB/sec 1363.81 MB/sec 2140.72 MB/sec 1255.08 MB/sec 1175.29 MB/sec 1084.75 MB/sec 1044.92 MB/sec 883.35 MB/sec 406.97 MB/sec 760.91 MB/sec 352.66 MB/sec

Notes

Following commands were used to test the board:

1 Thread:  sysbench --num-threads=1 --test=cpu --cpu-max-prime=2000 run
2 Threads: sysbench --num-threads=2 --test=cpu --cpu-max-prime=2000 run
4 Threads: sysbench --num-threads=4 --test=cpu --cpu-max-prime=2000 run
256M: sysbench --test=memory --memory-block-size=1M --memory-total-size=256M run
512M: sysbench --test=memory --memory-block-size=1M --memory-total-size=512M run

Conclusions

For new installations it is strongly recommend investing in multi-core (at least 2 cores) hardware, as single-core mini boards may have problems dealing with intensive traffic. If in addition to OGN receiver software you plan to run other software on the same hardware (e.g. you are active feeder of ADS-B data to FR24), then definitely do not consider any single-core solution.

If, however, you already have one or you can get one really cheap - for now we recommend to go for Raspberry Pi! because of these reasons:

  1. OGN receiver software is able to use GPU on Raspberry Pi (and that significantly improves performance, although e.g. BeagleBone Black without the use of GPU proves to offer similar performance as Pi with GPU)
  2. The Raspberry community is by far the biggest community. If you have a question or a special problem, then the probability is very high you get a solution if you use a Raspberry Pi
Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License